De novo DNA methyltransferases Dnmt3a and Dnmt3b ... - Nature

Oncogene (2005) 24, 3091–3099

& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc

De novo DNA methyltransferases Dnmt3a and Dnmt3b primarily mediate the cytotoxic effect of 5-aza-20 -deoxycytidine Masahiro Oka1, Amy M Meacham1, Takashi Hamazaki1, Nemanja Rodic´1, Lung-Ji Chang2 and Naohiro Terada*,1 1 Department of Pathology, University of Florida College of Medicine, PO Box 100275, 1600 SW Archer Rd, Gainesville, FL 32610, USA; 2Department of Molecular Genetics and Microbiology, Powell Gene Therapy Center and McKnight Brain Institute, University of Florida, Gainesville, FL 32610, USA

The deoxycytidine analog 5-aza-20 -deoxycitidine (5-azadC) is a potent chemotherapeutic agent effective against selective types of cancer. The molecular mechanism by which 5-aza-dC induces cancer cell death, however, is not fully understood. It has been accepted that the mechanism of toxicity is due to the covalent binding between the DNA methyltransferase (Dnmt) and 5-aza-dC-substituted DNA. In order to define which member of the Dnmt family plays a dominant role in the cytotoxicity, we examined the effect of 5-aza-dC on cell growth and apoptosis in various Dnmt null mutant embryonic stem (ES) cells. Of interest, Dnmt3a–Dnmt3b double null ES cells were highly resistant to 5-aza-dC when compared to wild type, Dnmt3a null, Dnmt3b null, or Dnmt1 null ES cells. The cellular sensitivity to 5-aza-dC correlated well with the expression status of Dnmt3 in both undifferentiated and differentiated ES cells. When exogenous Dnmt3a or Dnmt3b was expressed in double null ES cells, the sensitivity to 5-aza-dC was partially restored. These results suggest that the cytotoxic effect of 5-aza-dC may be mediated primarily through Dnmt3a and Dnmt3b de novo DNA methyltransferases. Further, the ability to form Dnmt-DNA adducts was similar in Dnmt1 and Dnmt3, and the expression level of Dnmt3 was not higher than that of Dnmt1 in ES cells. Therefore, Dnmt3-DNA adducts may be more effective for inducing apoptosis than Dnmt1-DNA adducts. These results imply a therapeutic potential of 5-aza-dC to cancers expressing Dnmt3. Oncogene (2005) 24, 3091–3099. doi:10.1038/sj.onc.1208540 Published online 21 February 2005 Keywords: DNA methyltransferase; azacytidine; embryonic stem cell; differentiation; apoptosis

Introduction Aberrant gene expression in cancer cells is generated by mutations in genomic DNA, chromosomal transloca*Correspondence: N Terada; E-mail: [email protected]fl.edu Received 29 November 2004; revised 13 January 2005; accepted 13 January 2005; published online 21 February 2005

tions, and frequently by abnormal epigenetic regulations (Laird and Jaenisch, 1996). DNA methylation in mammals is known as one of the most common epigenetic modifications involved in carcinogenesis. In general, cancer cells show genome-wide hypomethylation in addition to regional hypermethylation at CpG islands of particular genes (Baylin et al., 1998; Costello and Plass, 2001). For example, genome-wide hypomethylation was demonstrated to cause chromosomal instability, which eventually increases the incidence of T cell lymphoma in mice (Trinh et al., 2002; Eden et al., 2003; Gaudet et al., 2003). On the contrary, aberrant promoter hypermethylation associated with the silencing of tumor suppressor genes enhances carcinogenesis (Baylin et al., 1998; Jones and Laird, 1999). Despite the accumulated evidence of abnormal DNA methylation patterns in cancer cells, the molecular mechanisms causing them remain unrevealed. The DNA methyltransferase (Dnmt) family of proteins is responsible for the attachment of a methyl group to the 5-position of the cytosine ring when it is followed by a guanosine residue (Bestor, 2000). In mammals, global DNA methylation is catalysed mainly by three DNA methyltransferases, Dnmt1, Dnmt3a, and Dnmt3b. Dnmt1 has a high preference for hemimethylated DNA and is essential for maintaining methyla tion patterns during DNA replication, and thus is called the maintenance DNA methyltransferase (Leonhardt et al., 1992; Pradhan et al., 1999). Dnmt3a and Dnmt3b, on the other hand, are responsible for the wave of de novo methylation during early embryogenesis, which establishes the somatic methylation pattern (Okano et al., 1999). It has been reported that Dnmt1 is highly expressed in various cancer cells (el-Deiry et al., 1991; Issa et al., 1993; Belinsky et al., 1996; Melki et al., 1998). Interestingly, notable increases in expression of de novo DNA methyltransferase are also reported in diverse cancer cells and cell lines (Robertson et al., 1999; Xie et al., 1999; Kanai et al., 2001; Mizuno et al., 2001; Saito et al., 2001; Chen et al., 2002; Girault et al., 2003). It has also been demonstrated that sustained expression of Dnmt3a or Dnmt3b de novo DNA methyltransferase is necessary for embryonic stem (ES) cells to form teratomas in nude mice (Chen et al., 2003). Therefore, abnormal expression of

Dnmt3 mediates 5-aza-20 -deoxycytidine cytotoxicity M Oka et al

3092

de novo DNA methyltransferases may serve as a marker for cancer cells, as well as a potential target for future cancer therapy. The nucleoside analogue 5-aza-20 -deoxycytidine (5aza-dC), which has been widely used as a DNA demethylation agent (Jones and Taylor, 1980), is an effective chemotherapeutic agent for acute myelogenous leukemia (AML), chronic myeloid leukemia (CML) and advanced myelodysplastic syndromes (Gabbara and Bhagwat, 1995; Santini et al., 2001; Goffin and Eisenhauer, 2002). 5-aza-dC becomes incorporated into genomic DNA, and the resulting 5-aza-dC-substituted DNA forms adducts with DNA methyltransferase (Santi et al., 1984; Juttermann et al., 1994; Gabbara and Bhagwat, 1995; Ferguson et al., 1997; Christman, 2002). This covalent and irreversible binding of the enzyme to drug-substituted DNA, rather than secondary DNA demethylation due to enzyme depletion, is believed to be the principal cause for cytotoxicity (Michalowsky and Jones, 1987; Juttermann et al., 1994). In fact, a previous study demonstrated that Dnmt1 null cells were more resistant to 5-aza-dC than wild-type cells (Juttermann et al., 1994). However, it has not been addressed if de novo DNA methyltransferases, Dnmt3a, and Dnmt3b, are involved in the cellular responses to 5-aza-dC. In this paper, we examined the involvement of Dnmt3a and Dnmt3b in cellular toxicity of 5-aza-dC.

Results De novo DNA methyltransferase null ES cells are more resistant to 5-aza-dC compared to wild type or maintenance DNA methyltransferase null ES cells In ES cells, de novo methyltransferase Dnmt3a and Dnmt3b are expressed along with maintenance methyltransferase Dnmt1, contributing to the establishment and maintenance of genomic methylation patterns (Chen et al., 2003). To define which member of Dnmts plays a dominant role in the cytotoxic effect of 5-aza-dC, we examined the effect of 5-aza-dC on cell growth of wild type (WT), Dnmt1 null (Dnmt1/), and Dnmt3a– Dnmt3b double null (Dnmt3a/ Dnmt3b/) ES cells when cultured in ES maintenance medium containing LIF (ES-LIF medium) (Figure 1a). Compared to wildtype ES cells, Dnmt3a/ Dnmt3b/ ES cells showed about a 200-fold increase in resistance to 5-aza-dC (Figure 1b). In contrast, Dnmt1/ ES cells showed only about a four-fold increase in resistance. We then tested the effect of 5-aza-dC when ES cells were differentiated into embryoid bodies (EB) (Figure 1c). At day 2 of EB formation, wild type and Dnmt1/ ES cells showed a decrease in the size of EB in the presence of 5-aza-dC (0.1–1 mM), whereas Dnmt3a/ Dnmt3b/ ES cells were more resistant to 5-aza-dC in this culture condition as well. In contrast to 5-aza-dC sensitivity, Dnmt3a/ Dnmt3b/ ES cells were slightly more sensitive to TSA, a histone deacetylase inhibitor, than wild type or Dnmt1 null ES cells. At day 5 of the Oncogene

Figure 1 Dnmt3a–Dnmt3b double null ES cells showed resistance to 5-aza-dC. (a) Immunoblotting was performed to verify the expression of Dnmts in knockout ES cell lines. Cell lysate (25 mg) from wild type (WT), Dnmt1 null (Dnmt1/), and Dnmt3a– Dnmt3b double null (Dnmt3a/ Dnmt3b/) ES cells were analysed by immunoblotting with anti-Dnmt1, anti-Dnmt3, or antib-actin antibody. Undifferentiated ES cells (WT) express a high level of Dnmt3a2 compared to Dnmt3a1 and 3b1 (see Figure 6b). (b) Wild type (WT) ES cells, Dnmt1 null ES (Dnmt1/), and Dnmt3a–Dnmt3b double null ES (Dnmt3a/ Dnmt3b/) were plated at a density of 5 103 ES cells/well on a 24-well gelatincoated plate in the presence of LIF and various concentrations (0– 2 mM) of 5-aza-dC. After 5 days, cells were fixed with 3.7% formaldehyde for 15 min at room temperature, washed with PBS, and stained with methylene blue (Fisher Scientific). (c) Wild type, Dnmt1/ and Dnmt3a/ Dnmt3b/ ES cells were cultured for 2 days in differentiation medium by the hanging drop method in the presence of 5-aza-dC (0.1–1 mM) or TSA (300 nM). After 2 days, EBs in hanging drops were transferred to tissue culture dishes and observed. Bar, 1 mm


3093

differentiation, most of wild type and Dnmt1/ ES cells were dead when cultured in the presence of 1 mM of 5-aza-dC. In contrast, Dnmt3a/ Dnmt3b/ EB attached to the plate and could further proliferate (data not shown). These results suggest that expression of de novo DNA methyltransferase may be more responsible for the 5-aza-dC cytotoxicity than maintenance DNA methyltrnasferase in ES cells regardless of their differentiation status. Expression of either Dnmt3a or Dnmt3b de novo methyltransferase is sufficient for the cellular sensitivity to 5-aza-dC To examine which de novo DNA methyltransferase is primarily responsible for the 5-aza-dC cytotoxicity, we further compared wild type, Dnmt3a/, Dnmt3b/, and Dnmt3a/ Dnmt3b/ ES cells for 5-aza-dC sensitivity (Figure 2). In contrast to Dnmt3a/ Dnmt3b/ double null ES cells, Dnmt3a/ or Dnmt3b/ ES cells showed only a slight increase in resistance to 5-aza-dC compared to wild-type ES cells when cultured in ES-LIF medium (see Discussion). Similar results were obtained when cells were cultured in differentiation medium without LIF. In contrast, upon treatment with araC, a nucleotide analogue used for cancer treatment, all the ES cell lines showed a comparable sensitivity. Cisplatin, another chemotherapeutic agent known to form adducts with DNA, also had a similar cytotoxic effect on all the cell lines examined. Similarly, all cell lines demonstrated a comparable sensitivity to TSA (less than 10-fold differences). These results suggest that expression of either Dnmt3a or Dnmt3b de novo DNA methyl-

transferase alone is sufficient for the sensitivity to 5-aza-dC in ES cells. Expression of de novo DNA methyltransferase plays an important role in apoptotic cell death induced by 5-aza-dC treatment Since 5-azacytidine has been shown to induce apoptotic cell death (Gorczyca et al., 1993; Kizaki et al., 1993; Murakami et al., 1995), we examined the effect of 5-azadC on apoptotic cell death of ES cells with disrupted Dnmt alleles. ES cells on gelatin-coated dishes in ESLIF medium were treated with 5-aza-dC for 24 h. The appearance of the sub-G1 population, the cells with hypodiploid DNA that corresponds to apoptotic cells, was analysed by propidium iodide staining and subsequent flow cytometric analysis. As shown in Figure 3a (top), the sub-G1 population of wild-type ES cells increased by 10 mM 5-aza-dC treatment. However, using a relatively low dosage of 5-aza-dC (0.01–1 mM), which is sufficient to see the inhibitory effect on colony growth on plate (Figure 1b), did not cause an increase in sub-G1 cells within 24 h (data not shown). Treatment of the cells with 100 mM 5-aza-dC caused further increase in the subG1 population of cells, concomitant with the accumulation of cells in the G2/M phase of the cell cycle. In contrast, Dnmt3a/ Dnmt3b/ ES cells showed only a slight increase in the sub-G1 population even when they were treated with 100 mM 5-aza-dC for 24 h. The effect of 5-aza-dC on apoptotic cell death was also examined when ES cells were differentiated (Figure 3a, bottom). ES cells were differentiated for 2 days in the absence of LIF and treated with 5-aza-dC for an additional 24 h. Compared to undifferentiated ES cells,

Figure 2 Expression of either Dnmt3a or Dnmt3b is sufficient for the 5-aza-dC sensitivity in ES cells. Wild-type (WT) ES cells, Dnmt3a null ES (Dnmt3a/), Dnmt3a–Dnmt3b double null ES (Dnmt3a/ Dnmt3b/), and Dnmt3b null ES (Dnmt3b/) were plated at a density of 5 103 ES cells/well on a 24-well plate in ES maintenance medium or differentiation medium in the presence of 5-aza-dC (0–1 mM), AraC (0–100 mM), Cisplatin (0–100 mM), or TSA (0–300 nM). Cells not killed by the drug treatment were visualized by methylene blue staining as described above Oncogene


3094

important for the induction of apoptosis by 5-aza-dC treatment in ES cells. Exogenous expression of Dnmt3a or Dnmt3b restored the sensitivity to 5-aza-dC in Dnmt3a–Dnmt3b double null ES cells

Figure 3 Apoptosis induced by 5-aza-dC in de novo methyltransferase null ES cells. (a) 5-aza-dC increased the sub-G1 population in ES cells. (Top) Indicated ES cell lines were plated at a density of 5 104 cells/well on gelatin-coated six-well plate and cultured in ES maintenance medium. After 2 days, a final concentration of 10 or 100 mM of 5-aza-dC was added to the medium and incubated for an additional 24 h. Cells were harvested, stained with propidium iodide, and analysed by flow cytometry as described in Materials and methods. (Bottom) Indicated ES cell lines were plated at a density of 5 104 cells/well on gelatin-coated six-well plate and differentiated in the absence of LIF. Cell cycle was analysed as described above. (b) DNA fragmentation assay. Genomic DNA was extracted from 5-aza-dC-treated ES cells and separated on 1.5% agarose gels

differentiating ES cells showed substantial increase in sub-G1 population after 5-aza-dC treatment. Incubation with 100 mM 5-aza-dC caused apoptotic cell death for most of the differentiating wild-type ES cells (89%). However, much fewer sub-G1 cells (28%) were observed with Dnmt3a/ Dnmt3b/ ES cells under the same condition. Apoptotic cell death was further confirmed by DNA fragmentation assay (Figure 3b). 5-aza-dC treatment caused an increase in the fragmented DNA in wild-type cells both in undifferentiated or differentiated culture conditions. In contrast, we detected only a slight increase in the fragmented DNA in Dnmt3a/ Dnmt3b/ ES cells. Taken together, these results suggest that expression of Dnmt3a or Dnmt3b is Oncogene

To further test the hypothesis that expression of Dnmt3a or Dnmt3b is crucial in 5-aza-dC-mediated cytotoxicity, we reconstituted Dnmt3a or Dnmt3b expression in the double null ES cells. We used a lentiviral vector, which can escape gene silencing after stable integration into the host ES cell genome, thereby promoting more efficient protein expression (Hamaguchi et al., 2000; Pfeifer et al., 2002). Undifferentiated Dnmt3a/ Dnmt3b/ ES cells were infected with the lentivirus containing FLAGtagged Dnmt3a2 or Dnmt3b1 gene under the control of the constitutive EF1-a promoter. Dnmt3a2 and Dnmt3b1 are the splice variants that are most abundantly expressed among each subtype in ES cells (Chen et al., 2003). After infection, single colonies were isolated for the clonal expression of FLAG-Dnmt3a or FLAGDnmt3b. As shown in Figure 4a, cloned cells show uniform expression of FLAG-Dnmt3a or FLAGDnmt3b. However, we could not isolate a clone that shows high expression of Dnmt3a or Dnmt3b, most likely due to the cytotoxicity of overexpressed protein. Immunoblotting analysis revealed that the level of FLAG-Dnmt3a or FLAG-Dnmt3b is lower than endogenous Dnmt3a or Dnmt3b in wild-type ES cells (Figure 4b). These cells were plated and analysed for 5aza-dC sensitivity (Figure 4c). Compared to parental Dnmt3a/ Dnmt3b/ ES cells, Dnmt3a (clone 3A#10) or Dnmt3b (clone 3B#7) reconstituted cells partially restored the sensitivity to 5-aza-dC. DNA content analysis revealed that these Dnmt3a expressing clone 3A#10 or Dnmt3b expressing clone 3B#7 showed more sub-G1 populations of cells compared to parental Dnmt3a–Dnmt3b double null ES cells when they were treated with 5-aza-dC (Figure 4d). Similar results were obtained with additional clones each expressing FLAGDnmt3a or Dnmt3b. In addition, expression of a truncated Dnmt3b protein, deleting its catalytic domain, did not restore the sensitivity to 5-aza-dC (data not shown). These results further demonstrated that the expression of active de novo DNA methyltransferases Dnmt3a and Dnmt3b are critical for the sensitivity to 5-aza-dC. Similar ability of Dnmt-DNA adduct formation between Dnmt1 and Dnmt3 It has been accepted that the mechanism of 5-aza-dC cytotoxicity is due to the covalent binding between Dnmt and 5-aza-dC-substituted DNA. Therefore, the differential binding ability of Dnmt1 and Dnmt3 to 5-aza-dC-substituted DNA may explain the differential sensitivity to the drug. The Dnmt-DNA adduct formation was evaluated by the solubility of Dnmt in the cell lysate after 5-aza-dC treatment. The covalent binding of Dnmt to 5-aza-dC


3095

Figure 4 Expression of exogenous Dnmt3a or Dnmt3b restored the sensitivity to 5-aza-dC (a) Dnmt3a–Dnmt3b double null ES cells (Dnmt3a/ Dnmt3b/) were infected with lentiviral vector for exogenous FLAG-Dnmt3a or FLAG-Dnmt3b expression. Single clones were isolated for FLAG-Dnmt3a (Dnmt3a clone3A#10) or FLAG-Dnmt3b (Dnmt3b clone3B#7) after infection. Dnmt3a/ Dnmt3b/ ES cells, Dnmt3a expressing clone3A#10, and Dnmt3b expressing clone3B#7 were differentiated for 4 days and stained with anti-FLAG antibody. (b) Immunoblotting analysis of the level of reconstituted FLAG-Dnmt3a or FLAG-Dnmt3b. Total cell lysate (30mg) from day 4 EBs (WT, Dnmt3a/ Dnmt3b/, Dnmt3a clone3A#10, Dnmt3b clone3B#7) was used for immunoblotting for anti-Dnmt3, anti-FLAG and anti-b-actin. FLAG-tagged Dnmt3a2 and 3b1 showed a slower mobility when compared to endogenous proteins. (c) Infected cells were tested for their 5-aza-dC sensitivity as shown in Figure 1b. (d) Apoptosis of infected cells after 5-aza-dC treatment was examined by flow cytometry analysis as shown in Figure 3a

substituted DNA causes sequestration of soluble Dnmt into the insoluble fraction. As shown in Figure 5, the amount of soluble Dnmt1 decreased following 5-azadC treatment in a dose-dependent manner. The amount of soluble Dnmt3a or Dnmt3b also decreased, but less efficiently than Dnmt1 after the 5-aza-dC treatment. Thus, the degree of adduct formation of Dnmts unlikely explains their differential sensitivity to 5-aza-dC.

Protein expression levels of Dnmts in undifferentiated and differentiated ES cells Another explanation for why Dnmt3a and Dnmt3b play a more dominant role in 5-aza-dC sensitivity is that the protein expression level of Dnmt3a and Dnmt3b maybe higher than that of Dnmt1 in ES cells. If this is the case, there will be a greater chance for Dnmt3a and/or Dnmt3b to form Dnmt-DNA adducts than Dnmt1 does. Oncogene


3096

Figure 5 Depletion of Dnmts from soluble fraction after 5-aza-dC treatment. Wild-type ES cells were treated with 10 or 100 mM of 5aza-dC for 24 h. Cell lysates (30 mg) were examined for the level of soluble Dnmts. The same blot was deprobed and used for the detection of Dnmt1, Dnmt3, and b-actin

Since the antibody that shows equal affinity to both Dnmt1 and Dnmt3 are not available, we could not compare the protein expression level between Dnmt1 and Dnmt3 directly. To compare the level of Dnmt1 and Dnmt3, ES cell lysates were immunoblotted together with COS-7 cell lysates expressing FLAG-tagged Dnmt1 and Dnmt3a2 (Figure 6a). Transfected COS-7 cells showed comparable expression of FLAG-Dnmt1 and FLAG-Dnmt3a2. When compared to FLAG-Dnmt1 in COS-7 cells, the expression of endogenous Dnmt1 in undifferentiated ES cells was higher. In contrast, expression of endogenous Dnmt3a2 was lower than FLAG-Dnmt3a2 expressed in COS-7 cells. We can then conclude that the expression level of Dnmt1 was higher than that of Dnmt3a2 in ES cells. Accordingly, the difference in protein expression level between Dnmt1 and Dnmt3a or Dnmt3b may not explain the reason for the difference in the sensitivity to 5-aza-dC between undifferentiated Dnmt1/ and Dnmt3a/ Dnmt3b/ double null ES cells. Interestingly, the level of Dnmt3a and Dnmt3b were increased after ES cell differentiation, whereas the level of Dnmt1 was decreased. The bands for Dnmt3a and Dnmt3b were further confirmed by immunoblotting using knockout ES cells (Figure 6b). The expression of Dnmt3a1, Dnmt3a2, and Dnmt3b1 increased considerably after ES cell differentiation. These results support the idea that increased protein expression of Dnmt3a and Dnmt3b protein contributes to the hypersensitivity to 5-aza-dC during ES cell differentiation.

Discussion In this paper, we have demonstrated that Dnmt3a– Dnmt3b double null ES cells are more resistant to 5-azadC than wild-type ES cells and Dnmt1 null ES cells. In contrast, Dnmt3a or Dnmt3b single null ES cells showed only a slight increase in resistance to 5-aza-dC compared to wild-type ES cells. Furthermore, the ectopic expression of Dnmt3a or Dnmt3b in the double null ES cells Oncogene

Figure 6 Protein expression level of Dnmts in ES cells. (a) Expression of Dnmts was compared using FLAG-tagged Dnmts expressed in COS-7 cells as reference protein. Cell lysates (40 mg) from undifferentiated wild-type ES cells and differentiated ES (day 4 EB) were analysed by immunoblotting with lysate from COS-7 cell expressing FLAG-Dnmt1 and FLAG-Dnmt3. The same blot was used for anti-FLAG, anti-Dnmt1, anti-Dnmt3 anti-b-actin immunoblotting. (b) Dnmt3a and Dnmt3b expression in undifferentiated ES cells and differentiated ES (day 4 EB). Cell lysates (WT, Dnmt3a/, Dnmt3b/, Dnmt3a/ Dnmt3b/ ES) from undifferentiated ES cells or differentiated EBs (20 mg per lane) were analysed by immunoblotting with anti-Dnmt3 or anti-b-actin antibody

restored the sensitivity to 5-aza-dC. These results indicate that expression of de novo methyltransferase (either Dnmt3a or Dnmt3b) is critical for 5-aza-dC sensitivity in ES cells. A previous report showed that Dnmt1 null ES cells were more resistant to 5-aza-dC than wild-type ES cells (Juttermann et al., 1994), which was not inconsistent with our results. However, the present study revealed more dominant roles of Dnmt3a and Dnmt3b de novo methyltransferases in the cytotoxicity of 5-aza-dC. We also demonstrated that the expression level of Dnmt3 proteins change during ES cell differentiation (Figure 6), correlating with cellular sensitivity to 5-azadC. When ES cells were maintained undifferentiated in ES-LIF medium, Dnmt3a null ES cells showed more


3097

resistance to 5-aza-dC than Dnmt3b null ES cells (Figure 2), in agreement with a dominant expression of Dnmt3a2 over Dnmt3b1 in undifferentiated ES cells (Figure 6). Interestingly, the protein level of Dnmt3a and Dnmt3b were substantially increased after ES cell differentiation (Figure 6a and b), which associated with the considerable increase in 5-aza-dC sensitivity after ES dell differentiation (Figure 3). To identify the mechanism behind the dominant role of Dnmt3a and Dnmt3b in 5-aza-dC sensitivity to ES cells, we compared the protein level and adduct formation ability between Dnmt1 and Dnmt3. Immunoblotting analysis revealed that the protein expression level of Dnmt1 is higher than that of Dnmt3 in ES cells (Figure 6a), contradicting our hypothesis that protein expression level was the cause. Further, the adduct formation patterns between Dnmt1 and Dnmt3 (Liu et al., 2003; Weisenberger et al., 2004) refuted the idea that the affinity of Dnmt3 for 5-aza-dC must be higher in order to cause sensitivity. It is still possible, however, that the difference in the quality of Dnmt-DNA adducts may delineate the phenomenon. Of interest, the protein localization of Dnmt1 and Dnmt3 in the nucleus is known to be different. Dnmt1 localizes to replication foci during S phase to maintain the DNA methylation pattern during DNA replication (Leonhardt et al., 1992; Chuang et al., 1997). Therefore, Dnmt1-DNA adducts may be mainly formed at the replication foci, where DNA excision repair is highly active. Interestingly, Dnmt1 associates with proliferating cell nuclear antigen (PCNA) (Chuang et al., 1997), which is essential for DNA repair (Shivji et al., 1992; Umar et al., 1996). In contrast, Dnmt3a and Dnmt3b prefer to localize at pericentric heterochromatin (Bachman et al., 2001), where the Dnmt-DNA adducts might be more stable once it is formed. Therefore, Dnmt1-DNA adducts might be more easily repaired than Dnmt3a or Dnmt3b-DNA adducts. The adducts may be recognized as abnormal DNA that causes activation of a downstream signaling pathway. In fact, it has been previously shown that the p53 pathway is activated after 5-aza-dC treatment as a DNA damage response (Karpf et al., 2001; Nieto et al., 2004; Zhu et al., 2004). The Dnmt3-DNA adducts may be more effective for inducing apoptosis than Dnmt1-DNA adducts. During ES cell differentiation, the transcriptional activity of p53 is upregulated (Lin et al., 2005), despite the downregulation of p53 protein levels (Sabapathy et al., 1997). Therefore, active p53 might promote hypersensitivity to 5-aza-dC during ES cell differentiation in addition to the increased expression of de novo methyltransferases. Indeed, the expression of p53 is proposed to play an important role in 5-aza-dC sensitivity in cancer cell lines (Karpf et al., 2001; Schneider-Stock et al., 2005). In contrast, Nieto et al. (2004) have shown that p53 null mouse embryonic fibroblast (MEF) cells are more sensitive to 5-aza-dC than p53 positive MEF cells due to the defect in cell-cycle checkpoint arrest. Therefore, the downstream effect of 5-aza-dC-activeted p53 might

be dependent on cell lines. It is interesting to see if p53 positively or negatively affects 5-aza-dC sensitivity in ES cells. By using a lentiviral vector for the ectopic expression of Dnmt3a or Dnmt3b, we successfully increased the sensitivity of double-knockout cells to 5-aza-dC. Dnmt3a and Dnmt3b are known to be expressed at relatively low levels in fully differentiated cells and adult tissues (Okano et al., 1998; Xie et al., 1999). In contrast, Dnmt1 is expressed ubiquitously in most human tissues (Robertson et al., 1999). Given that overexpression of de novo DNA methyltransferase is occasionally observed in some cancer cells (Robertson et al., 1999; Xie et al., 1999; Kanai et al., 2001; Mizuno et al., 2001; Saito et al., 2001; Chen et al., 2002; Girault et al., 2003), 5-aza-dC may be more effective in selective types of cancer cells in which Dnmt3 expression is upregulated. Indeed, 5-azadC treatment has been effective for CML, AML, and advanced myelodysplastic syndrome (Santini et al., 2001; Goffin and Eisenhauer, 2002). Interestingly, expression of Dnmt3a and Dnmt3b as well as Dnmt1 is upregulated in AML and CML cells (Mizuno et al., 2001). The expression of Dnmt3b has been demonstrated to be required for the proliferation and survival of certain human cancer cells (Beaulieu et al., 2002), and contributes to oncogenic transformation (Soejima et al., 2003). These results raise the possibility that many types of cancer cells may be expressing Dnmt3b and could be potential targets for 5-aza-dC treatment. In summary, we have demonstrated that Dnmt3a and Dnmt3b de novo methyltransferase predominantly mediates the cytotoxic effect of 5-aza-dC, thus implying a potential application of this drug in treating cancers that expresses de novo methyltransferases. Materials and methods Cell culture and drug treatment Mouse ES cell lines J1 (wild type), Dnmt1 null (Dnmt1/), Dnmt3a null (Dnmt3a/), Dnmt3b null (Dnmt3b/), and Dnmt3a–Dnmt3b double null (Dnmt3a/ Dnmt3b/) ES cells were gifts from Dr En Li (Massachusetts General Hospital, Harvard Medical School). ES cells were maintained on gelatin-coated tissue culture dishes in a Dulbecco’s modified Eagle medium (DMEM) optimized for ES cells (Invitrogen) containing 1000 U/ml recombinant mouse leukemia inhibitory factor (LIF (ESGRO); Chemicon International), 10% knockout serum replacement (KSR; Invitrogen), 1% fetal calf serum (Atlanta Biologicals), 20 mM HEPES (Invitrogen), 300 mM monothioglycerol (Sigma), 100 U/ml penicillin, 100 mg/ml streptomycin (Invitrogen), and 2 mM Lglutamine (Invitrogen). In vitro ES cell differentiation was induced in an Iscove’s modified Dulbecco’s medium (Invitrogen) containing 20% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin, 100 mg/ml streptomycin, 300 mM monothioglycerol. The embryoid body formation was induced in a hanging drop containing approximately 2000 cells in differentiation medium for 2 days. COS-7 cells were obtained from American Type Culture Collection, and cultured in DMEM supplemented with 10% fetal calf serum, 100 U/ml penicillin, 100 mg/ml streptomycin. For drug sensitivity testing, freshly prepared 5-aza-20 -deoxycytidine (Sigma), cytosine Oncogene


3098 arabinofuranoside (araC; Sigma), cis-diamminedichloroplatinum(II) (cisplatin; Sigma), and trichostatin A (TSA; Wako) were added to culture media to the indicated final concentrations.

using ECL Western blotting detection reagents (Amersham Pharmacia).

DNA content analysis

Undifferentiated ES cells or differentiated ES cells (day 2) were treated with various concentrations of 5-aza-dC for 24 h. The cells were collected, and genomic DNA was extracted using Wizard Genomic DNA Purification Kit (Promega). In total, 2.5 mg of total genomic DNA for each sample was electrophoresed in 1.5% agarose gel and visualized by ethidium bromide staining.

Lentiviral vector, pTYF-EF, was used to construct FLAGDnmt3a2 and FLAG-Dnmt3b1 expression vectors under the control of the EF1-a promoter (Zaiss et al., 2002). Briefly, oligonucleotides encoding the FLAG-tag sequence were initially inserted into the BamHI site of pTYF-EF (pTYFEF-FLAG). Sequentially, the Dnmt3a2 or the Dnmt3b1 coding region sequence was inserted into the downstream site for amino terminally FLAG-tagged fusion protein. The constructed plasmid was packaged into lentivirus by cotransfection with pNHP (a helper plasmid) and pHEF-VSVG (an envelope coding plasmid) (Iwakuma et al., 1999). Transduction of the viral vectors was performed as described previously (Iwakuma et al., 1999). Expression of the FLAG-tagged proteins was confirmed both by immunoblotting and immunostaining. Furthermore, DNA methyltransferase activity of FLAG-Dnmt3a2 or FLAG-Dnmt3b1 was confirmed using MSRF (Methylation Sensitive Restriction Fingerprinting) analysis (data not shown). To isolate single clones, transducted cells were plated at low density in 100 mm dishes, and colonies were picked. At least 12 clones were isolated for each construct and tested for uniform expression of FLAG-Dnmt3a2 or FLAG-Dnmt3b1.

Immunoblotting analysis

Immunoflourescence analysis

Cells were collected by centrifugation and cell pellets were washed with PBS and resuspended in a radioimmunoprecipitation assay buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS) supplemented with protease inhibitors (leupeptin, aprotinin, and phenylmethylsulfonyl fluoride) and incubated on ice for 20 min. After centrifugation at 16 000 g for 5 min at 41C, supernatants were transferred to fresh tubes and used for further analysis. Protein extracts were then separated on 6.5% polyacrylamide gel by electrophoresis and transferred onto nitrocellulose membrane. After blocking with 4% bovine serum albumin (BSA) solution in TBST buffer (50 mM Tris-HCl (pH 8.0), 100 mM NaCl, 0.1% Tween 20) for 1 h at room temperature, the membrane was incubated with primary antibodies, 64B1466 monoclonal anti-Dnmt3a antibody (1 : 250 dilution; Imgenex), polyclonal anti-Dnmt1 antibody (1 : 5000 dilution) (Takagi et al., 1995; Suetake et al., 2001), or C-11 polyclonal antiactin antibody (1 : 2000 dilution; Santa Cruz Biotechnology, Inc.) for 2 h at room temperature. After incubation with secondary antibodies conjugated to horseradish peroxidase (1 : 5000 dilution; ICN), detection of horseradish peroxidase activity was performed

ES cells were cultured on gelatin-coated plates, washed once with PBS, and fixed in 3.7% formaldehyde/PBS for 15 min at room temperature. Cells were then permeabilized with 0.5% Triton X-100/PBS for 5 min and blocked with 5% BSA/PBS for 2 h at room temperature. Cells were further incubated with anti-FLAG M2 antibody (1 : 1000 dilution; Stratagene) for 2 h at room temperature. After four washes with PBS, cells were incubated with anti-mouse IgG conjugated to tetramethylrhodamine isothiocyanate (TRITC; 1 : 200 dilution; Jackson Immunoresearch Laboratories). After four washes with PBS, cells were mounted by Vectashield containing 4,6 diamidino-2phenylindole (Vector Laboratories). Fluorescent images were observed under an inverted microscope (IX-70, Olympus/C Squared), and captured by using MagnaFire digital camera system (Optronics).

DNA content analysis was performed by flow cytometry. Cells were harvested by centrifugation, washed in ice-cold phosphate-buffered saline (PBS), and fixed in 70% ethanol that had been prechilled to 201C. They were then resuspended in PBS and treated with RNase A (final concentration of 250 mg/ml; Sigma), and further incubated with PBS containing propidium iodide (final concentration of 50 mg/ml; Sigma) for 30 min on ice. Cell cycle analysis was performed using 20 000 cells with a FACSort flow cytometer using the Cellquest analysis program (Becton Dickinson). The data were further analysed using the Modifit software (Becton Dickinson). DNA fragmentation analysis

Lentiviral vector construction and transduction

Acknowledgements We thank Dr En Li for providing Dnmt null ES cells, Dr Shoji Tajima for anti-Dnmt1 antibody, and Dr Keith D Robertson and Dr Thomas C Rowe for critical reading of the manuscript.

References Bachman KE, Rountree MR and Baylin SB. (2001). J. Biol. Chem., 276, 32282–32287. Baylin SB, Herman JG, Graff JR, Vertino PM and Issa JP. (1998). Adv. Cancer Res., 72, 141–196. Beaulieu N, Morin S, Chute IC, Robert MF, Nguyen H and MacLeod AR. (2002). J. Biol. Chem., 277, 28176–28181. Belinsky SA, Nikula KJ, Baylin SB and Issa JP. (1996). Proc. Natl. Acad. Sci. USA, 93, 4045–4050. Bestor TH. (2000). Hum. Mol. Genet., 9, 2395–2402. Chen T, Ueda Y, Dodge JE, Wang Z and Li E. (2003). Mol. Cell. Biol., 23, 5594–5605. Oncogene

Chen T, Ueda Y, Xie S and Li E. (2002). J. Biol. Chem., 277, 38746–38754. Christman JK. (2002). Oncogene, 21, 5483–5495. Chuang LS, Ian HI, Koh TW, Ng HH, Xu G and Li BF. (1997). Science, 277, 1996–2000. Costello JF and Plass C. (2001). J. Med. Genet., 38, 285–303. Eden A, Gaudet F, Waghmare A and Jaenisch R. (2003). Science, 300, 455. el-Deiry WS, Nelkin BD, Celano P, Yen RW, Falco JP, Hamilton SR and Baylin SB. (1991). Proc. Natl. Acad. Sci. USA, 88, 3470–3474.


3099 Ferguson AT, Vertino PM, Spitzner JR, Baylin SB, Muller MT and Davidson NE. (1997). J. Biol. Chem., 272, 32260–32266. Gabbara S and Bhagwat AS. (1995). Biochem. J., 307 (Part 1), 87–92. Gaudet F, Hodgson JG, Eden A, Jackson-Grusby L, Dausman J, Gray JW, Leonhardt H and Jaenisch R. (2003). Science, 300, 489–492. Girault I, Tozlu S, Lidereau R and Bieche I. (2003). Clin. Cancer Res., 9, 4415–4422. Goffin J and Eisenhauer E. (2002). Ann. Oncol., 13, 1699–1716. Gorczyca W, Gong J, Ardelt B, Traganos F and Darzynkiewicz Z. (1993). Cancer Res., 53, 3186–3192. Hamaguchi I, Woods NB, Panagopoulos I, Andersson E, Mikkola H, Fahlman C, Zufferey R, Carlsson L, Trono D and Karlsson S. (2000). J. Virol., 74, 10778–10784. Issa JP, Vertino PM, Wu J, Sazawal S, Celano P, Nelkin BD, Hamilton SR and Baylin SB. (1993). J. Natl. Cancer Inst., 85, 1235–1240. Iwakuma T, Cui Y and Chang LJ. (1999). Virology, 261, 120–132. Jones PA and Laird PW. (1999). Nat. Genet., 21, 163–167. Jones PA and Taylor SM. (1980). Cell, 20, 85–93. Juttermann R, Li E and Jaenisch R. (1994). Proc. Natl. Acad. Sci. USA, 91, 11797–11801. Kanai Y, Ushijima S, Kondo Y, Nakanishi Y and Hirohashi S. (2001). Int. J. Cancer, 91, 205–212. Karpf AR, Moore BC, Ririe TO and Jones DA. (2001). Mol. Pharmacol., 59, 751–757. Kizaki H, Ohnishi Y, Azuma Y, Mizuno Y and Ohsaka F. (1993). Immunopharmacology, 25, 19–27. Laird PW and Jaenisch R. (1996). Annu. Rev. Genet., 30, 441–464. Leonhardt H, Page AW, Weier HU and Bestor TH. (1992). Cell, 71, 865–873. Lin T, Chao C, Saito S, Mazur SJ, Murphy ME, Appella E and Xu Y. (2005). Nat. Cell Biol., 7, 165–171. Liu K, Wang YF, Cantemir C and Muller MT. (2003). Mol. Cell. Biol., 23, 2709–2719. Melki JR, Warnecke P, Vincent PC and Clark SJ. (1998). Leukemia, 12, 311–316. Michalowsky LA and Jones PA. (1987). Mol. Cell. Biol., 7, 3076–3083. Mizuno S, Chijiwa T, Okamura T, Akashi K, Fukumaki Y, Niho Y and Sasaki H. (2001). Blood, 97, 1172–1179. Murakami T, Li X, Gong J, Bhatia U, Traganos F and Darzynkiewicz Z. (1995). Cancer Res., 55, 3093–3098.

Nieto M, Samper E, Fraga MF, Gonzalez de Buitrago G, Esteller M and Serrano M. (2004). Oncogene, 23, 735–743. Okano M, Bell DW, Haber DA and Li E. (1999). Cell, 99, 247–257. Okano M, Xie S and Li E. (1998). Nat. Genet., 19, 219–220. Pfeifer A, Ikawa M, Dayn Y and Verma IM. (2002). Proc. Natl. Acad. Sci. USA, 99, 2140–2145. Pradhan S, Bacolla A, Wells RD and Roberts RJ. (1999). J. Biol. Chem., 274, 33002–33010. Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J, Gonzales FA and Jones PA. (1999). Nucleic Acids Res., 27, 2291–2298. Sabapathy K, Klemm M, Jaenisch R and Wagner EF. (1997). EMBO J., 16, 6217–6229. Saito Y, Kanai Y, Sakamoto M, Saito H, Ishii H and Hirohashi S. (2001). Hepatology, 33, 561–568. Santi DV, Norment A and Garrett CE. (1984). Proc. Natl. Acad. Sci. USA, 81, 6993–6997. Santini V, Kantarjian HM and Issa JP. (2001). Ann. Intern. Med., 134, 573–586. Schneider-Stock R, Diab-Asseff M, Rohrbeck A, FoltzerJourdainne C, Boltze C, Hartig R, Schonfeld P, Roessner A and Gali-Muhtasib H. (2005). J. Pharmacol. Exp. Ther., 312, 525–536. Shivji KK, Kenny MK and Wood RD. (1992). Cell, 69, 367–374. Soejima K, Fang W and Rollins BJ. (2003). Oncogene, 22, 4723–4733. Suetake L, Shi L, Watanabe D, Nakamura M and Tajima S. (2001). Cell Struct. Funct., 26, 79–86. Takagi H, Tajima S and Asano A. (1995). Eur. J. Biochem., 231, 282–291. Trinh BN, Long TI, Nickel AE, Shibata D and Laird PW. (2002). Mol. Cell. Biol., 22, 2906–2917. Umar A, Buermeyer AB, Simon JA, Thomas DC, Clark AB, Liskay RM and Kunkel TA. (1996). Cell, 87, 65–73. Weisenberger DJ, Velicescu M, Cheng JC, Gonzales FA, Liang G and Jones PA. (2004). Mol. Cancer Res., 2, 62–72. Xie S, Wang Z, Okano M, Nogami M, Li Y, He WW, Okumura K and Li E. (1999). Gene, 236, 87–95. Zaiss AK, Son S and Chang LJ. (2002). J. Virol., 76, 7209–7219. Zhu WG, Hileman T, Ke Y, Wang P, Lu S, Duan W, Dai Z, Tong T, Villalona-Calero MA, Plass C and Otterson GA. (2004). J. Biol. Chem., 279, 15161–15166.

Oncogene